Revisiting Neural Activation Coverage for Uncertainty Estimation

2026-04-24

AI Summary

This work addresses the challenge of effectively estimating predictive uncertainty for pre-trained regression neural networks without requiring retraining. It extends Neural Activation Coverage (NAC)—originally developed for out-of-distribution detection in classification tasks—to regression settings for the first time, introducing a coverage-based uncertainty scoring mechanism. The proposed method systematically leverages activation patterns to quantify prediction confidence and is rigorously compared against established approaches such as Monte Carlo Dropout. Experimental results demonstrate that the NAC-based uncertainty scores significantly outperform existing methods in both discriminative capability and semantic meaningfulness, thereby establishing a new avenue for uncertainty quantification in regression models.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Uncertainty Representations
• Natural Language Processing: Interpretability, Analysis, and Evaluation of NLP Models

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Web Mining and Content Analysis: Large pretrained models with web data
• Social Networks and Social Media: Influence propagation, information diffusion, and the prediction on networks

Abstract

Neural activation coverage (NAC) is a recently-proposed technique for out-of-distribution detection and generalization. We build upon this promising foundation and extend the method to work as an uncertainty estimation technique for already-trained artificial neural networks in the domain of regression. Our experiments confirm NAC uncertainty scores to be more meaningful than other techniques, e.g. Monte-Carlo Dropout.

Problem

• Neural Activation Coverage
• Uncertainty Estimation
• Regression
• Out-of-Distribution Detection
• Artificial Neural Networks

Innovation

• Neural Activation Coverage
• Uncertainty Estimation
• Out-of-Distribution Detection
• Regression
• Monte-Carlo Dropout